About the paper:
“Evolution of Gender in Indo–European Languages”

Note: this page is part of the author’s set of pages on his research in cognitive science.

In the fall of 2001 I started writing a paper on the evolution of gender in Indo–European languages, which was published in the proceedings of the annual Cognitive Science Conference in August 2002.

If you are interested in the published version of the paper, here it is (in pdf):

But if you are interested in learning what the paper is about, please read the “concrete”, below. (My “concretes” explain in down-to-earth terms the contents of my papers, as opposed to the paper’s “abstract”, the purpose of which is to befuddle the reader, and convince the reviewing committee that the author knows how to speak “academese” properly.)


I was motivated to write this paper as a graduate student after listening to my advisor’s suggestion to read a paper by Lera Boroditsky and Lauren A. Schmidt, titled “Sex, Syntax, and Semantics”. My advisor has a keen interest in sexist language (among a myriad of other keen interests), and believes that many Indo–European languages, other than English, are hopelessly sexist. By this he means (I hope I am not misrepresenting his ideas) that the grammatical gender is so much interwoven with the nouns and pronouns in such languages, that to produce a gender-free (“non-sexist”) version of some phrase is often impossible. For example, in English we can say “the doctor”, and this is a non-sexist description of a person, because it does not tell us whether the doctor is a man or a woman. But in Spanish, for example, we have to say “el médico” if the doctor is a man, or “la médica” if it’s a woman; we have to reveal the doctor’s gender (or sex — the two words, gender and sex, can be used interchangeably in English, and this is a central idea here); there is no way to remain neutral in Spanish.

My advisor asked me to read the Boroditsky & Schmidt paper because he thought it was supportive of some of his ideas. In other words, he liked it. I read it, but disliked it. The first thing that predisposed me negatively was its sexy title: “Sex, Syntax, and Semantics”, when in fact there was not any mention of syntax in the paper. Obviously the title was made up so as to be catchy, with the three S’s and its mild metric pattern (1-2-3 syllables). I perceived that as a cheap trick: if you have something to say, say it in the content of your paper, don’t try to trap the reader with a title that sounds “cool” but (part of it at least) is irrelevant to your ideas. But a second, and far more important reason, was that I disagreed with the content of the paper. Here is why:

Boroditsky & Schmidt claim somewhere in their paper that native speakers of Indo–European languages such as French, Italian, Spanish, German, etc., which have formal gender,(1) think in male/female terms (and occasionally neuter, if the language allows it) when they think of objects, because nouns for objects in such languages have a grammatical gender which is masculine or feminine (or neuter). For example, when the Spanish speaker says “la puerta” (and the Italian “la porta”, and the Greek “η πόρτα”, and the Russian “дверь”, etc.) for “the door”, they activate the concept “female” (and maybe “woman”) in their minds, because the “door” in those languages is of feminine gender. Likewise for masculine nouns.

Fine. So far I had no great objection, because this sounds reasonable, even if uninteresting. I mean, “la puerta” is of feminine gender, the word “feminine” is related to “female” (in those languages, too), “female” to “woman”, so one thing brings another. Psychologists have shown with clever experiments that when we think of a concept its related concepts are primed in our minds, meaning that we don’t necessarily think “woman!” consciously, but subconsciously we activate the concept. (And don’t hurry to object that you don’t feel “woman” activated while saying “la puerta” if you are a Spanish native speaker, because the subconscious access cannot be felt introspectively.) I say this is uninteresting because it is well-known: the concept “door” activates subconsciously various other concepts, such as “wood” (the material out of which doors are usually made), hence “tree”, and even “forest”, “rectangle”, “open/close”, “floor” (by mere rhyming similarity with “door”), and a host of other concepts, one of which is “woman”. Naturally, the more distant a concept is from the original, the less activation it receives.

But Boroditsky & Schmidt said something more interesting. They claimed that people perceive an inherent masculinity or femininity in objects, independently of the language they speak, and for this reason, Boroditsky & Schmidt made the following two interesting predictions: (1) that Indo-European languages should show an agreement in gender assignments to nouns, and (2) that speakers of English who had no prior exposure to foreign languages, when asked to assign a masculine or feminine gender to nouns, they should agree not only among themselves, but also with the gender assignments of other Indo-European languages that have a formal gender (such as Spanish, German, etc.).

So I set out to examine these predictions. For prediction (1), what I found is that Indo-European languages indeed show some agreement in gender assignments, but this can be explained not by appeal to some supposed universal perception of gender in objects among humans,(2) but by the evolution of languages in time: if we assume that there existed an ancestor “Proto-Indo-European” language which assigned genders to objects (arbitrarily from our point of view, but rationally from its speakers’ point of view, perhaps due to their animistic conceptualizations of the world), and that this language evolved and bifurcated again and again, giving the tree of relationships of common origin that we observe at present in Indo-European languages, then the assignment of gender to nouns evolved similarly and changed arbitrarily (because there is no inherent and objective reason for assigning masculinity or femininity to objects, contrary to Boroditsky & Schmidt’s claim). This is exactly what I observed: for example, Spanish and Portuguese show a very close agreement in their assignments of gender to nouns because they are both Romance languages, relatively recent descendants of a common descendant of Latin; but Spanish and Dutch show a more remote agreement, because their common ancestor is more ancient, and whereas Spanish is a Romance language, Dutch is a Germanic language. Similarly, Spanish and Russian (a Slavic language) show an even more remote agreement, because their common ancestor language dates from even further back in the past; and so on, with this pattern observed in all pairs of 14 languages that I examined (Albanian, Dutch, French, German, Greek, Icelandic, Irish, Italian, Kurdish, Polish, Portuguese, Russian, Serbo-Croatian, and Spanish).

As for prediction (2), it fell flat on its face: according to the data collected, although monolingual native speakers of English (American college students) agree quite well among themselves in the way they assign masculinity or femininity to nouns (objects, abstract ideas, animals, etc.), they disagree with just about every other speaker of Indo-European languages in the world. To agree among themselves is expected, because they share a common culture, so when you ask them to think of a noun as masculine or feminine they make an association of masculinity with strength, robustness, large size, and so on, so they say that things like a hammer, pistol, boulder, thunder, anger, power, war, revolution, attack, dog, wolf, lion, bear, etc., are masculine; in contrast, by associating femininity with charm, beauty, small size, calmness, fragility, etc., they say that things like rain, flower, love, friendship, happiness, sadness, cat, butterfly, bird, etc., are feminine. But, as is explained in the paper, such choices conflict with all other I-E languages, showing only the faintest hint of correlation — a result that does not allow us to conclude that there is anything like a universally perceived masculinity or femininity in objects.

Further details are given in the paper. Enjoy!



Footnotes (clicking on the number of the footnote brings back to the text):

(1) Languages have formal gender if they assign a gender to nouns arbitrarily. For example, in Greek the door is feminine, the wall is masculine, and the floor is neuter. Languages such as English are said to have natural gender, because inanimate objects are almost always assigned to the neuter gender (“it”), whereas animate males are assigned to the masculine (“he”) and animate females to the feminine gender (“she”). (There are only minor exceptions and inconsistencies, such as using “she” for a boat, ship, or country; using “he” for certain animals of unknown sex, such as a bear, a wolf, etc.)

(2) After all, the notion of “gender” is more general than the way it appears in Indo-European languages: other languages of the world make completely different categories, such as putting all plants in one category, all metal objects in another, all divine entities in yet another one, and so on, and each of these categories constitutes a gender — this is briefly reviewed in the introductory section of the paper. So it cannot be that humans have some universal perception of masculinity and femininity.


Back to Harry’s index to research topics